A compilation of aircraft observations of the atmospheric surface layer over Denmark Strait, the Iceland Sea, and the Irminger Sea are compared against several meteorological analyses and QuikSCAT wind products. The observations are during cold-air outbreak conditions and moderate to high wind speeds during February and March 2007. About 150 data points spread over six days are used, with each data point derived from a 2-minute run (equivalent to a 12 km spatial average). The observations were taken 30-50 m above the sea surface and are adjusted to standard meteorological heights for the comparison. Surface-layer temperature, humidity, wind speed and direction, as well as sea surface temperature and surface turbulent fluxes are compared against co-located data from the ECMWF operational analyses (at two truncations); NCEP Global Reanalyses; NCEP North American Regional Reanalyses; Met Office NAE (North Atlantic and European) operational analyses; two MM5 hindcasts; and two QuikSCAT wind products.

In general, the limited area models are better at capturing the mesoscale high wind speed features and their associated structure. A high-resolution prescription of the sea surface temperature field is crucial for accurate surface-layer temperature, humidity and surface fluxes in this region, although this was not generally used at this time in these models. The most significant discrepancies a poor simulation of relative humidity in the NCEP Global and MM5 models; a cold bias in 2-m air temperature near the sea-ice edge in the NAE model; and an overestimation of wind speed above 20 m s-1 in the QuikSCAT wind products, particularly the RSS (Remote Sensing System) retrieval. In addition, the NCEP Global, NARR and MM5 models all have significant discrepancies associated with the parameterization of surface turbulent heat fluxes.